Behavioral The official journal of the

Ecology ISBE
International Society for Behavioral Ecology

Behavioral Ecology (2015), 26(6), 1470–1475. doi:10.1093/beheco/arv089

Original Article

Human perception of fighting ability: facial

Editor’s choice

cues predict winners and losers in mixed

martial arts fights
Anthony C. Little,a Vít Třebický,b Jan Havlíček,c S. Craig Roberts,a and Karel Kleisnerb

Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015

aDivision of Psychology, School of Natural Sciences, University of Stirling, Stirling FK9 4LA, UK and
bDepartment of Philosophy and History of Sciences and cDepartment of Zoology, Faculty of Sciences,

Charles University, Viničná 7, Prague, Czech Republic

Received 18 December 2014; revised 14 May 2015; accepted 22 May 2015; Advance Access publication 7 July 2015.

In antagonistic encounters, the primary decision to be made is to fight or not. Animals may then possess adaptations to assess fight-
ing ability in their opponents. Previous studies suggest that humans can assess strength and fighting ability based on facial appear-
ance. Here we extend these findings to specific contests by examining the perception of male faces from paired winners and losers
of individual fights in mixed martial arts sporting competitions. Observers, unfamiliar with the outcome, were presented with image
pairs and asked to choose which of the 2 men was more likely to win if they fought while other observers chose between the faces
based on masculinity, strength, aggressiveness, and attractiveness. We found that individuals performed at rates above chance in cor-
rectly selecting the winner as more likely to win the fight than the loser. We also found that winners were seen to be more masculine,
stronger, and more aggressive than losers. Finally, women saw the winners as more attractive than the losers. Together these findings
demonstrate that 1) humans can predict the outcome of specific fighting contests based on facial cues, 2) perceived masculinity and
strength are putative cues to fighting success available from faces, and 3) facial cues associated with successful male–male competi-
tion are attractive to women.
Key words:  competition, face appearance, fighting, intrasexual, violence.

INTRODUCTION assessing fighting ability in their opponents (Parker 1974; Enquist

Across many animal species, fighting as a form of intrasexual selec-
tion (competition between members of the same sex) is common
and has led to the evolution of animal weapons, such as horns and
antlers, particularly in males (Andersson 1994). Adaptive decisions,
or fitness-enhancing decisions, rely on balancing the net benefits
against the net costs of particular actions (Krebs and Davies 1998).
In antagonistic encounters with other individuals of the same spe-
cies, the primary decision to be made is to fight or not. The ben-
efits to be gained, such as territory, must be weighed against the
costs, the potential for injury or even death.
Although the benefits of fighting will vary across species and
environment, the same costs are applicable to many species, and
critically, the costs vary greatly depending on whether an animal is
likely to be the winner or loser of the fight. We can then expect that
animals that engage in intraspecies fighting will possess perceptual/
cognitive adaptations to assess the risks involved in this behavior by
Address correspondence to A.C. Little. E-mail: anthony.little@stir.ac.uk.
and Leimar 1983) using cues that are potentially related to fight-
ing ability such as body size, strength, and weaponry (Krebs and
Davies 1998). Indeed, there is evidence that animals make deci-
sions about fighting based on the assessment of the relative fighting
abilities of their opponents (Gosling et al. 1996; Hazlett 1996) and
that specific traits in some species can be related to fighting success.
For example, in terms of visual perception, variable black facial
patterns in paper wasps are related to both body size and social
dominance (Tibbetts and Dale 2004) and red chest coloration in
gelada baboons is related to troop status, with leader males hav-
ing the reddest chests (Bergman et al. 2009). Individuals could base
their decisions to fight on appearance-linked cues to fighting ability
allowing them to compete when likely to win and to avoid costly
agonistic interactions when likely to lose.
In humans, there is evidence that male–male competition is
important across various different cultures. For example, as noted
by Sell et  al. (2009), fighting ability is associated with access to
resources in the Yanomamo of Venezuela (Chagnon 1983), the
Achuar of Ecuador (Patton 2000), and the Tsimane of Bolivia (von

© The Author 2015. Published by Oxford University Press on behalf of

the International Society for Behavioral Ecology. All rights reserved. For
permissions, please e-mail: journals.permissions@oup.com
Little et al. • Facial cues to fighting success
Rueden et  al. 2008). In other cultures, sports involving ritualized
combat between men are common and take many forms, such as
Sumo in Japan and stick-fighting in the Suri of Ethiopia. These
ritualized forms of combat have a long recorded history, includ-
ing fencing in the 16th century Germany and gladiatorial combat
in Ancient Rome. In line with this evidence for physical combat
between men, also noted by Sell et  al. (2009, 2012), there are a
range of anatomical and physiological sex differences that appear
to reflect adaptation to male–male competition in humans, includ-
ing sex differences in height and upper body strength (Plavcan and
Van Schaik 1997; Puts 2010).
Given evidence for intrasexual conflict in humans and follow-
ing theoretical predictions for adaptations to assess fighting abil-
ity (Parker 1974; Enquist and Leimar 1983), previous researchers
have suggested that humans possess adaptations to infer fighting
ability, specifically that fighting ability might be inferred from
facial, body, and vocal cues (Sell et  al. 2009, 2010). For exam-
ple, people make relatively accurate inferences about men’s
physical strength from static facial images (Sell et  al. 2009) and
voice recordings (Sell et  al. 2010), and measurements of physi-
cal strength are associated with ratings of fighting ability (Sell
et  al. 2009). One study has shown that self-rated fighting ability
is positively related to acquaintance-rated fighting ability, which
in turn is positively related to unfamiliar-person-rated fighting
ability based on face photographs (Doll et al. 2014). This work is
suggestive of cues to fighting ability being available in faces, but
it is important to note that self-ratings and acquaintance ratings
are likely to be noisy measures of real fighting ability. Focusing
on human facial cues, masculinity in male faces has been associ-
ated with perceived dominance (Perrett et  al. 1998) and physical
strength is positively related to ratings of facial masculinity (Fink
et  al. 2007). Recent studies have also highlighted that face mea-
surements are associated with aggression in men. For example,
facial width scaled for face height is correlated with perceived
aggression (Carré et al. 2009), related to self-reported dominance
and, relating to real behavior, aggressive behavior in sport (Carre
and McCormick 2008; Třebický et al. 2015; Zilioli et al. 2015).
Further, one study examining forensic data from skeletons has
shown that men with narrow faces are more likely to have died
from contact violence than their wider faced peers (Stirrat et  al.
2012).
While the accurate assessment of strength and its association
with fighting ability (Sell et  al. 2009), links between facial mea-
surements and aggression (Carre and McCormick 2008), and that
studies have associated fighting success with facial measurements
showing that men with wider faces relative to height are more
likely to win in mixed martial arts (MMA) competition (Třebický
et al. 2015; Zilioli et al. 2015) are all in line with the notion that
humans can assess fighting ability from facial cues, they do not
provide direct evidence for this notion. One study has, however,
examined fighting success based on instances of real fights in
MMA sporting contests. Calculating fighting success as the ratio of
wins to losses across a fighter’s Ultimate Fighting Championship
(UFC) fighting career, it was found that the perceived aggressive-
ness of fighters’ faces was linked to their success in actual physical
confrontations, although perceived fighting ability and differences
in facial shape were only associated with fighting success in heavy-
weight fighters (Třebický et al. 2013). This suggests that perceived
aggression may be an underlying cue to fighting success rather
than the cognitively complex inferred fighting success. However,
fighting success across fights is somewhat different to assessing
1471
fighting outcomes from faces in particular contests between pairs
of fighters. In other words, only one face is relevant when assess-
ing general fighting ability, whereas, in specific contests, individu-
als can compare the traits of 2 protagonists. This comparison
may enable greater accuracy in judgment. Being able to predict
the outcome of contests between 2 individuals may be adaptive
because it allows for discrimination between individuals within a
group in order to select successful allies or mates. The cue used to
discriminate between pairs of others could also be used to assess
a person’s relative fighting ability. For example, an individual may
be able to compare their own estimated ability to a competitor’s
ability based on appearance to predict their own chances of suc-
cessfully winning a fight.
In the current study, we examined individual’s abilities to directly
assess the outcome of particular fights. Although previous results
suggest that individuals can assess the fighting ability of particu-
lar fighters from their faces based on their overall success across
a number of fights (Třebický et  al. 2013), here we focused on a
Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015
more fine-grained analysis in which face images of fighters were
presented as pairs such that observers were tasked to judge the
difference in perceived traits of the winners and losers of specific
fights. We asked observers to judge between the winners and losers
of fights for a variety of traits to test ideas relating to intrasexual
and intersexual selection. First, we addressed accuracy in judg-
ment by asking observers to choose who they think would win in a
fight. Accuracy at this level would indicate that observers are able
to assess the relative fighting ability of 2 fighters to correctly predict
the outcome. Second, we examined specific cues from faces that
may underlie accuracy: perceived masculinity, strength, and aggres-
siveness. Third, we addressed attractiveness to the opposite-sex
because, while perception of fighting ability is often considered the
domain of intrasexual selection, it may also be related to intersex-
ual selection. In terms of attractiveness to the opposite-sex, there
are benefits that could be associated with preferring better fighters:
1) indirect benefits, genetic benefits that are passed to offspring such
as genes associated with health, strength, or high quality immune
systems and 2)  direct benefits, benefits that are directly passed to
mates or offspring such as resources or protection from other males.
We then also asked a sample of women who they thought was more
attractive out of the pair.
METHODS
Participants acting as observers
There were 5 different studies in which participants chose between
pairs of faces for different traits. Independent groups of partici-
pants judged between faces for: who would win in a physical fight
(N = 69, men = 32, women = 37, mean age = 29.7, standard devia-
tion [SD] = 10.7, 95% confidence interval [CI], lower: 27.1, upper:
32.2), who is more masculine (N  =  33, men  =  11, women  =  22,
mean age  =  25.6, SD  =  8.1, 95% CI, lower: 22.7, upper: 28.5),
who is stronger (N = 30, men = 10, women = 20, mean age = 30.3,
SD = 12.7, 95% CI, lower: 25.5, upper: 35.0), who is more aggres-
sive (N = 30, men = 12, women = 18, mean age = 27.4, SD = 8.2,
95% CI, lower: 24.3, upper: 30.4), and who is more attractive
(N  =  34, women  =  34, mean age  =  29.0, SD  =  11.3, 95% CI,
lower: 25.1, upper: 33.0). Participants were selected for being older
than 16  years of age. For attractiveness judgments, only women
reporting to be heterosexual were selected for analysis. Participants
were recruited for the study online via a research-based website and
the study was conducted online.
1472
Stimuli
The original study population consisted of 285 MMA fighters for
which facial photographs and details of their previous fight (oppo-
nent and win/loss), as well as facial photographs of their opponent,
were available from the official Web site of the MMA division of
the UFC (www.ufc.com; database accessed in June 2012). Because
this represented the total pool of fighters, excepting unselected
fighters for which data or photographs were unavailable, it was pos-
sible to match the 285 fighters with their opponent in their most
recent fight. Out of the 285 fighters, we created 156 pairs of fights
based on the most recent matches for the fighters. From these pairs,
42 pairs were excluded from the analyses because they contained a
duplicate fighter from one of the preceding fight pairs.
The final set of images used were of 228 fighters which made up
114 unique pairs representing fights between 2 different fighters.
Using the available database, for each pair, 1 fighter was classified
as the winner and 1 as the loser.
For each pair of fighters, we obtained data on their weight class,
which was the same for each fighter making up the pair. To reduce
the number of classifications and increase the sample size of final
groupings, we averaged the 7 available weight classes into 3 group-
ings: lightweight (bantam weight, feather weight, light weight, N = 48
pairs), middleweight (welter weight, middle weight, N  =  42 pairs),
and heavyweight (light heavy weight, heavy weight, N = 24 pairs).
The stimulus set comprised the official front-on photographs
available from www.ufc.com. These photographs appear to have
approximately similar lighting and background with individuals
posing with an approximately neutral expression. To equate size of
the face in the image, all images were aligned to standardize the
position of the pupils in the image.
Procedure
Participants were administered a short questionnaire assessing age,
sex, and sexual orientation (only used for women rating attractive-
ness), followed by a forced-choice face test. There were 5 different
forced-choice face tests for which the stimuli and procedure were
identical except that participants in each test were given different
instructions on what type of discrimination they were asked to do.
Different participants took part in each of the tests based on ran-
dom allocation to tests.
In the forced-choice tests, the 114 pairs of winners and losers of
MMA fights as described above were shown with both order and
side of presentation randomized. Participants were asked to choose
1 face from the pair for a particular trait. Clicking a button below
the face selected moved participants on to the next face trial. There
was no time limit for responses and both faces remained on screen
until participants selected a face.
Specific questions for the 5 tests were:
“Which person is more likely to WIN in a physical fight?”
“Which person is more MASCULINE?
“Which person is PHYSICALLY STRONGER?”
“Which person is more AGGRESSIVE?”
“Which person is more ATTRACTIVE?”
Statistical analyses
The dependent variable was the choice by each participant of the
winner or loser for each pair of fighters for 114 pairs. If the partici-
pant selected the winner from the pair, this was scored “1” and if
the participant selected the loser from the pair, this was scored “0”.
Behavioral Ecology
First, general linear mixed model (GLMM), or multilevel model-
ing, analyses were conducted using R (R Core Team 2013); specifi-
cally, we used the “glmer” function available in the “lme4” package
(Bates et al. 2014). Such models allow simultaneous analysis of par-
ticipant and stimulus effects negating the need to collapse across
either. Participant (1|subject) and face pair (1|fight) were specified
as random factors in the model. The nature of data entered here
was binary (0/1), and so a binomial model was specified using the
“glmer” function which fitted the model using maximum likelihood
with Laplace approximation. The model as specified in R was as
follows:
  pickwinner ~ (1|subject ) + (1| fight ) ,  
 modelA = glmer  data = fight , family = binomial  

In this model, where “pickwinner” was whether the subject cor-
rectly chose the winner, we tested for a significant effect of the
intercept which would indicate a difference from chance (0).
Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015
A second model was specified in which sex of participant and
weight category were added as fixed effects to the above model:
  pickwinner ~ (1|subject ) + (1| fight ) +  
 modelB = glmer  sexparticipant + weightcategory , 
  data = fight , family = binomial  
 
Models were compared using the “Anova” function. A  nonsignifi-
cant difference between models would indicate that adding sex of
participant and weight category did not impact significantly on the
original model.
To follow-up these analyses, we additionally included a by-partic-
ipant and by-face analysis using 1-sample t-tests to test if choice of
winner over loser was significantly different from chance. Impact of
weight category was tested in the by-face analysis using Anova and
impact of sex of participant was tested in the by-participant analy-
sis using independent samples t-tests.
RESULTS
General linear mixed models
Separate models were computed for: who would win in a physical
fight, who is more masculine, who is stronger, who is more aggres-
sive, and who is more attractive.
The first model indicated that choice of “Which person is more
likely to WIN in a physical fight” was a significant predictor of win-
ning a match (Z = 2.35, P = 0.019). Adding sex and weight category
to the model created a model that was not significantly different
from the original model (chi square = 1.66, df = 3, P = 0.645).
The second model indicated that choice of “Which person is
more MASCULINE?” was a significant predictor of winning a
match (Z = 2.00, P = 0.038). Adding sex and weight category to the
model created a model that was not significantly different from the
original (chi square = 2.54, df = 3, P = 0.469).
The third model indicated that choice of “Which person is
PHYSICALLY STRONGER?” was a significant predictor of win-
ning a match (Z = 2.00, P = 0.045). Adding sex and weight category
to the model created a model that was not significantly different
from the original for strength (chi square = 2.27, df = 3, P = 0.518).
The fourth model indicated that choice of “Which person is
more AGGRESSIVE?” was a significant predictor of winning a
match (Z = 2.57, P = 0.010). Adding sex and weight category to the
Little et al. • Facial cues to fighting success 1473

model created a model that was not significantly different from the
original for aggressiveness (chi square = 6.17, df = 3, P = 0.104).
The last model indicated that choice of “Which person is more
ATTRACTIVE?” was a nonsignificant predictor of winning a
match (Z  =  1.76, P  =  0.079), although the P value was close to
0.05. Adding weight category to the model created a model that
was not significantly different from the original for attractiveness
(chi square = 0.43, df = 2, P = 0.808).
In all of the above models, winners were selected more often
than losers. A summary of model statistics for each question is pre-
sented in Table 1.
To examine the equivalence of the GLMM analysis with meth-
ods involving calculation of means, because these types of analy-
sis are common in the literature, we carried out further analyses in
which mean choice was calculated for each face pair and for each
participant. We note that variance across fighters is most important
to the question of whether individual fighter’s faces contain cues
to fighting success and so the GLMM above and the by-face pair
separately for: who would win in a fight, who is more mascu-
line, who is stronger, and who is more attractive. We additionally
tested for effects of sex of participant using independent samples
t-tests.
One-sample t-tests indicated that winners were chosen sig-
nificantly more often than losers for winning in a physical
fight (t68  =  7.86, P  <  0.001, D  =  1.91), being more masculine
(t32  =  4.93, P  <  0.001, D  =  1.74), being stronger (t29  =  6.57,
P  <  0.001, D  =  2.44), being more aggressive (t29  =  5.34,
P  <  0.001, D  =  1.98), and being more attractive (t33  =  6.96,
P < 0.001, D = 2.42).
Independent samples t-tests indicated no significant effect of
sex of participant for judgments of winning in a physical fight
(t67 = 0.69, P = 0.493, D = 0.17), masculinity (t31 = 0.31, P = 0.762,
D = 0.11), strength (t28 = 1.46, P = 0.156, D = 0.55), or aggressive-
ness (t28 = 0.97, P = 0.342, D = 0.37).
Correlations among judgments

Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015

analyses are more appropriate to answer this question. Using data by face pair, we ran Pearson product-moment corre-
lations to examine relationships between the different attributes.
By face pair Correlations can be seen in Table  2. Significant positive correla-
Mean choice of winner versus loser was calculated for each face tions were found among the judgments of winning in a physical
pair and face pair was used as the unit of analysis and compared fight, masculinity, strength, and aggressiveness (all r > 0.490, all
with chance with 1-sample t-tests. This was done separately for: P  <  0.001). None of these variables, however, was significantly
who would win in a fight, who is more masculine, who is stron- related to attractiveness judgments (all P > 0.05).
ger, and who is more attractive. We additionally tested for effects of Previous authors have argued that masculinity in male faces
weight category using 1-way Anovas. may not be attractive because it is associated with negative attribu-
One-sample t-tests indicated that winners were chosen sig- tions, such as aggressiveness (Little et  al. 2011; Puts et  al. 2012).
nificantly more often than losers for winning in a physical We tested this idea by examining the relationship between choice
fight (t113  =  2.36, P  =  0.020, D  =  0.44), being more masculine as more masculine and choice as more attractive while control-
(t113  =  2.17, P  =  0.032, D  =  0.41), and being more aggressive ling for both choice as more aggressive and stronger. To examine
(t113 = 2.74, P = 0.007, D = 0.52). Although winners were chosen how women’s preferences were related to these traits indepen-
more often than losers, this was not significantly different from dently, we entered perceived masculinity, strength, and aggres-
chance for being stronger (t113  =  1.97, P  =  0.052, D  =  0.37) and sion as predictors of women’s attraction in a linear regression.
being more attractive (t113 = 1.71, P = 0.091, D = 0.32). This revealed a significant overall model (F3,110 = 6.23, P < 0.001,
One-way Anovas (dependent variable  =  mean choice of win-
ner, fixed factor  =  weight category) indicated no significant effect
of weight category for judgments of winning in a physical fight Table 2
(F2,111  =  0.72, P  =  0.491, ηp2  =  0.013), masculinity (F2,111  =  1.15, Intercorrelations among perceived traits based on the choice of
P  =  0.319, ηp2  =  0.020), strength (F2,111  =  0.32, P  =  0.724, a face out of a pair for each question
ηp2 = 0.006), aggressiveness (F2,111 = 2.37, P = 0.099, ηp2 = 0.041), Masculine Strong Aggressive Attractive
or attractiveness (F2,111 = 0.14, P = 0.871, ηp2 = 0.002).
Win fight 0.783** 0.815** 0.720** 0.150
By participant Masculine 0.743** 0.699** 0.175
Strong 0.490** 0.139
Mean choice of winner versus loser was calculated for each par- Aggressive −0.120
ticipant, and participant was used as the unit of analysis and
compared with chance with 1-sample t-tests. This was done **Significant at P < 0.01.

Table 1
Model summaries for choice of the winner as more likely to win, more masculine, stronger, more aggressive, and more attractive of
114 pairs of fighters
Winner Masculine Strong Aggressive Attractive

Estimate 0.203 0.183 0.238 0.238 0.193

Mean 0.550 0.546 0.559 0.559 0.548
Standard error 0.086 0.088 0.119 0.092 0.110
Z/P value 2.35/0.019 2.08/0.038 2.00/0.045 2.57/0.010 1.76/0.079

Participant, N 69 31 30 30 34

Estimate is the probability of picking the winner on the logit scale and the standard error reported is that of the estimate.
1474
R2  =  0.145) in which masculinity was significantly positively
(beta  =  0.514, P  <  0.001), aggressiveness was significantly nega-
tively (beta  =  −0.474, P  <  0.001), and physical strength was not
significantly (beta  =  −0.011, P  =  0.932) associated with women’s
choices for attractiveness.
DISCUSSION
Our data demonstrated that both men and women perceive win-
ners of fights differently from losers. Specifically, from the mixed
model analyses, winner’s faces were more likely to be seen as able
to win the fight, physically stronger, more aggressive, more mascu-
line, more aggressive, and more attractive to women than loser’s
faces (although this last effect was nonsignificant, P  =  0.079). We
found no significant effects of sex of observer or weight category
of fighter for these judgments. Similar effects were seen in by-
participant and by-face pair analyses, although effects were stron-
gest in the by-participant analyses. This difference is the result of
greater variance between face pairs than between observers in
terms of choices. For example, while the mean choice is identical
(M = 0.543), for choice of winner as winning in a physical fight, the
SD was lower across participant (SD = 0.05) than across face pair
(SD = 0.20).
Given the potential importance of male intrasexual selection
in human evolution (Chagnon 1983; Plavcan and Van Schaik
1997; Patton 2000; von Rueden et  al. 2008), our data are in line
with the notion that humans possess perceptual/cognitive adap-
tations to assess the risks involved fighting by assessing fighting
ability in other humans, as expected in a species that engages in
such behavior (Parker 1974; Enquist and Leimar 1983). Although
previous researchers have suggested that humans possess adap-
tations to detect fighting ability based on perceptions of strength
(Sell et al. 2009, 2010) and correspondence between self-rated and
acquaintance-rated fighting ability (Doll et al. 2014), here we show
direct evidence that humans can predict the actual outcome of
specific fights based on facial information, in line with a previous
demonstration that the perceived aggressiveness of fighters’ faces
was linked to their career fighting success (Třebický et  al. 2013).
Although humans do not necessarily have obvious evolved pheno-
typic weaponry, such as horns or antlers seen in nonhuman species
(Krebs and Davies 1998), humans may display cues to their fighting
abilities and possess adaptations to help guide their choice to fight
specific individuals (Parker 1974; Enquist and Leimar 1983).
We tested for sex differences in each judgment but found no
significant effects. It might be expected that men would pay more
attention to cues to male–male competitive ability because such
contests are more relevant to them, but our data suggests that
women perform similarly in discriminating winners from losers on
the basis of facial appearance (see also Třebický et al. 2015). We
note, however, that our sample sizes were relatively small for exam-
ining sex differences because this was not the main aim of the study.
Sex differences may indeed be found using larger sample sizes or
in alternative situations that emphasize the relevance to men over
women, such as in real-life competitive situations.
We note that across all types of judgments, the perceptual dif-
ference between winners and losers was relatively small. Given the
number of other variables that could determine the winner and
loser of these fights, we think it would be surprising if facial cues
accounted for the majority of the variance, and of course, small
advantages can prove important over evolutionary time scales.
There are also other reasons why the effects seen here are likely to
Behavioral Ecology
be modest. In our study, observers were limited to seeing static 2D
face information. Stronger relationships between facial appearance
and fight outcome may be possible under different experimental
conditions, for example, if participants were given 3D face images
or were exposed to the faces for more time. Given our interest was
in static facial cues, we excluded lots of potential cues to fighting
ability. In real-life fights, body size and dynamic cues are available
which may increase accuracy. Additionally, the fighters here belong
to a relatively homogenous group of highly trained athletes and are
therefore well matched. This is an interesting case in discriminating
winners and losers as this is likely to be a harder task than pre-
dicting who will win in less balanced fights. Indeed, fighters here
were also further matched in terms of weight category specifically
designed to create more even odds. In real fighting situations, where
weight, as a proxy for muscle mass or strength, is more uneven, we
might predict greater success in predicting the outcomes of fights.
In terms of specific cues to fighting success, winner’s faces were
generally seen as more masculine and stronger than loser’s faces.
Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015
Facial masculinity is then a potential cue to fighting ability and is also
positively related to perceived dominance (Perrett et  al. 1998), real
physical strength (Fink et al. 2007), and testosterone levels, although
the relationship with testosterone may be somewhat more complex
than a simple linear relationship (Pound et al. 2009). Judgments of
perceived physical strength from faces have been previously high-
lighted as a proxy for judgments of fighting ability (Sell et al. 2009),
with perceived strength relating to actual measured strength (Sell
et al. 2009). There are also links between facial measurements and
aggression (Carre and McCormick 2008) and one previous study has
shown that fighters with more aggressive appearing faces are more
likely to have higher success in their fights over the careers (Třebický
et al. 2013). Given these traits are potentially interlinked, they could
all relate to fighting success via the same mechanism. For example,
underlying levels of testosterone could underpin facial cues to mas-
culinity, strength and aggression. Of course these traits may be also
associated with fighting success for different reasons. For example,
strength may be a good predictor of who wins fights because it is
linked directly to the outcome of competition, but in more evenly
matched fights, cues to behavioral aggression may also be used to
predict winners independent of strength (see also Třebický et  al.
2013). In fact, there may be shared and unshared factors relating to
fighting success for each of these 3 factors.
In predicting women’s preferences, the zero-order correlations
indicated nonsignificant correlations between other judgments and
attractiveness judgments. However, when controlling for other judg-
ments in the regression analysis, masculinity was positively related,
aggressiveness negatively related, and strength was unrelated to
faces being selected as attractive to women. This is suggestive that
while women found winner’s faces as more attractive than losers,
this was due to differences in perceived masculinity. This further
highlights that masculinity and aggressiveness, while having similar
effects on perceived intrasexual competition abilities (winning fights),
have quite different effects in term of intersexual selection (their
attractiveness to women). Indeed, the benefits of avoiding aggres-
sive male partners are clear despite the fact that such males may
be successful in intrasexual competition. Here, controlling for per-
ceived aggressiveness and strength, the relationship between judg-
ments of masculinity and attractiveness increased from r  =  0.175
to 0.514 (Z  =  2.92, P  =  0.004). Previous studies have shown that
women moderate their preferences for masculine facial cues accord-
ing to their recent experience of visual environmental cues of direct
male–male competition and violence. In these previous studies,
Little et al. • Facial cues to fighting success
women preferred more masculine male faces after exposure to cues
of direct male–male competition and violence (Little et  al. 2013),
which is consistent with idea that women here preferred the faces of
men who were most likely to be successful in male–male competi-
tion. Perhaps such preferences reflect that ideal men should be able
to compete successfully but not actively seek out conflict (indicated
by high perceived aggression). In this way, women may select men
who can defend themselves, their partner, and their offspring from
other men but who do not continually seek conflict. Indeed, it has
previously been argued that women may face a trade-off in select-
ing masculine appearing partners because, while such partners may
be more dominant, masculine partners may not possess behavioral
traits, such as cooperativeness or faithfulness, that are desirable in a
long-term partner (Little et al. 2011; Puts et al. 2012). In such pref-
erences, it is difficult to tease apart the role of indirect from direct
benefits. This is because preferences for successful competition can
relate to both. For example, preferring men who are likely to win
in fights can lead to direct benefits in terms of resources as such
men may most successfully defend or acquire resources. However,
such preferences can also lead to potential indirect benefits by pass-
ing genes for the successful defense or acquisition of resources on
to male offspring, if these factors are heritable. In other words, if
women prefer traits that are associated with the ability to provide
direct benefits and mate with these men, the factors associated with
ability to provide direct benefits may also be passed to her offspring
thereby providing indirect benefits (see Kokko et al. 2003). It is then
likely that both direct and indirect benefits from men play a role in
generating preferences for the faces of men likely to win fights.
In summary, we found that individuals performed at rates
above chance in correctly selecting the winner as more likely to
win the fight than the loser. We also found that winners were
seen to be more masculine, more aggressive, and stronger than
losers. Finally, women saw the winners as more attractive than
the losers. The effect sizes for each of these relationships were
generally small but could have potentially important evolution-
ary consequences. Together these findings demonstrate that
1)  humans can correctly predict the outcome of specific fight-
ing contests, 2)  perceived masculinity/strength/aggressiveness
are all putative cues to fighting ability available from faces, and
3)  facial cues associated with successful male–male competition
are attractive to women.
FUNDING
A.C.L. was supported by a Royal Society University Research
Fellowship. V.T., J.H., and K.K. are supported by a Czech Science
Foundation grant (15-05048S).
Handling editor: Alexei Maklakov
REFERENCES
Andersson M. 1994. Sexual selection. Princeton (NJ): Princeton University
Press.
Bates D, Maechler M, Bolker B, Walker S. 2014. lme4: linear mixed-effects
models using Eigen and S4. R Package Version 1.1-7. http://CRAN.R-
project.org/package=lme4.
Bergman TJ, Ho L, Beehner JC. 2009. Chest color and social status in male
geladas (Theropithecus gelada). Int J Primatol. 30:791–806.
Carre JM, McCormick CM. 2008. In your face: facial metrics predict
aggressive behaviour in the laboratory and in varsity and professional
hockey players. Proc Biol Sci. 275:2651–2656.
1475
Carré JM, McCormick CM, Mondloch CJ. 2009. Facial structure is a reli-
able cue of aggressive behavior. Psychol Sci. 20:1194–1198.
Chagnon N. 1983. Yanomamo: the fierce people. 3rd ed. New York: Holt,
Rinehart and Winston.
Doll LM, Hill AK, Rotella MA, Cardenas RA, Welling LLM, Wheatley JR,
Puts DA. 2014. How well do men’s faces and voices index mate quality
and dominance? Hum Nat. 25:200–212.
Enquist M, Leimar O. 1983. Evolution of fighting behaviour: decision rules
and assessment of relative strength. J Theor Biol. 102:387–410.
Fink B, Neave N, Seydel H. 2007. Male facial appearance signals physical
strength to women. Am J Hum Biol. 19:82–87.
Gosling LM, Atkinson NW, Dunn S, Collins SA. 1996. The response of
subordinate male mice to scent marks varies in relation to their own com-
petitive ability. Anim Behav. 52:1185–1191.
Hazlett BA. 1996. Assessments during shell exchanges by the hermit crab
Clibanarius vittatus: the complete negotiator. Anim Behav. 51:567–573.
Kokko H, Brooks R, Jennions MD, Morley J. 2003. The evolution of mate
choice and mating biases. Proc Biol Sci. 270:653–664.
Krebs JR, Davies NB. 1998. Behavioural ecology: an evolutionary
approach. Oxford: Blackwell Science.
Little AC, DeBruine LM, Jones BC. 2013. Environment contingent pref-
erences: exposure to visual cues of direct male-male competition and
Downloaded from http://beheco.oxfordjournals.org/ by guest on November 23, 2015
wealth increase women’s preferences for masculinity in male faces. Evo
Hum Behav. 34:193–200.
Little AC, Jones BC, DeBruine LM. 2011. Facial attractiveness: evolution-
ary based research. Philos Trans R Soc B Biol Sci. 366:1638–1659.
Parker GA. 1974. Assessment strategy and evolution of fighting behavior. J
Theor Biol. 47:223–243.
Patton J. 2000. Reciprocal altruism and warfare: a case from the Ecuadorian
Amazon. In: Cronk L, Chagnon N, Irons W, editors. Adaptation and
human behavior: an anthropological perspective. Hawthorne (NY):
Aldine de Gruyter. p. 417–436.
Perrett DI, Lee KJ, Penton-Voak IS, Rowland DR, Yoshikawa S, Burt DM,
Henzi SP, Castles DL, Akamatsu S. 1998. Effects of sexual dimorphism
on facial attractiveness. Nature. 394:884–887.
Plavcan M, Van Schaik C. 1997. Interpreting hominid behavior on the
basis of sexual dimorphism. J Hum Evol. 32:345–374.
Pound N, Penton-Voak IS, Surridge AK. 2009. Testosterone responses
to competition in men are related to facial masculinity. Proc Biol Sci.
276:153–159.
Puts DA. 2010. Beauty and the beast: mechanisms of sexual selection in
humans. Evo Hum Behav. 31:157–175.
Puts DA, Jones BC, DeBruine LM. 2012. Sexual selection on human faces
and voices. J Sex Res. 49:227–243.
von Rueden C, Gurven M, Kaplan H. 2008. The multiple dimen-
sions of male social status in an Amazonian society. Evol Hum Behav.
29:402–415.
Sell A, Bryant GA, Cosmides L, Tooby J, Sznycer D, von Rueden C,
Krauss A, Gurven M. 2010. Adaptations in humans for assessing physical
strength from the voice. Proc Biol Sci. 277:3509–3518.
Sell A, Cosmides L, Tooby J, Sznycer D, von Rueden C, Gurven M. 2009.
Human adaptations for the visual assessment of strength and fighting
ability from the body and face. Proc Biol Sci. 276:575–584.
Sell A, Hone LSE, Pound N. 2012. The importance of physical strength to
human males. Hum Nat. 23:30–44.
Stirrat M, Stulp G, Pollet TV. 2012. Male facial width is associated with
death by contact violence: narrow-faced males are more likely to die from
contact violence. Evo Hum Behav. 33:551–556.
R Core Team. 2013. R: a language and environment for statistical comput-
ing. Vienna (Austria): R Foundation for Statistical Computing. Available
from: http://www.R-project.org/.
Tibbetts EA, Dale J. 2004. A socially enforced signal of quality in a paper
wasp. Nature. 432:218–222.
Třebický V, Fialová J, Kleisner K, Roberts SC, Little AC, Havlíček J. 2015.
Further evidence for links between facial width-to-height ratio and fight-
ing success: commentary on Zilioli et al. Aggress Behav. 41:331–334.
Třebický V, Havlícek J, Roberts SC, Little AC, Kleisner K. 2013. Perceived
aggressiveness predicts fighting performance in mixed-martial-arts fight-
ers. Psychol Sci. 24:1664–1672.
Zilioli S, Sell AN, Stirrat M, Jagor J, Vickerman W, Watson NV, Jagore
J. 2015. Face of a fighter: bizygomatic width as a cue of formidability.
Aggress Behav. 41:322–330.